Multistage Diagnosis of Alzheimer's Disease Based on Slice Attention Network

Xiaolai Huo, Chung-Ming Own, Yue Zhou, Nannan Wu, Jianwen Sun

Abstract: Alzheimer’s disease (AD) is a latent progressive neurodegenerative disease. Early detection can prevent further damage to patient’s health. We proposed a 3D abnormal perception depth residual network based on the squeeze and excitation module (RSE) and recurrent slice attention module (RSA). In our model, RSE captures the importance of different channels by integrating extrusion and excitation modules into residual blocks, while RSA aims to model 3D MRI images as slice sequences to capture the long-term dependence of different slices in different directions. Our model combine the context information of the abnormal area with local and spatial information. Experimental results show that the accuracy of our method is 87.5%, which is better than the most advanced model in terms of normal cognition (NC), early mild cognitive impairment (EMCI), late mild cognitive impairment (LMCI) and Alzheimer’s disease (AD) on the ADNI dataset. The CAM visualization results also show that our method can successfully highlight the most contributing regions of 3D MRI images.